Allow longer sleep or graceful kill for transfer container

[annybs]

In some environments I encounter errors like the following while managing nodes:

[critical] Error: (HTTP code 409) unexpected - Cannot kill container: b669ef9ad779e3bf312e430fa358ffb3b1f27ca5eec403f1896ebc5ecc5e2be4: Container b669ef9ad779e3bf312e430fa358ffb3b1f27ca5eec403f1896ebc5ecc5e2be4 is not running  {}

This appears to be related to slow disk access. We don't want the transfer container to stop prematurely in this case. A solution is needed to improve the reliability of device volume read.